Modeling of Malaria Prevalence in Indonesia with Geographically Weighted Regression

Abstract

Malaria is a public health problem that can lead to death, especially in high-risk groups i.e. infants, toddlers and pregnant women. This disease is still endemic in most parts of Indonesia. The relation of location factor between regions with the surrounding region was assumed to give the effect of spatial variability in the prevalence of malaria in the region. It would lead to the prevalence of malaria modeling using classical regression methods become less precise due to the assumption of homogeneity of variance was not met. It could be overcome by Geographically Weighted Regression (GWR) modeling. In GWR analysis, the selection weighting function was one determinant of the analysis results. GWR analysis resulted on the prevalence of malaria in Indonesia, GWR model with bisquare kernel weighting function had a better value of R² and AIC than GWR models with gaussian kernel weighting function.